Study on Text Classification Algorithm Based on Non-negative Matrix Factorization

In order to solve the high computational overhead and low classification efficiency of the KNN algorithm, a text feature vector representation method based on information gain and non-negative matrix factorization is proposed. The algorithm uses the information entropy and the semantic analysis method to realize the selection of the classification feature, which can reduce the number of text classification features, reducing the dimension of the text feature vector. Therefore, the overhead of similarity calculation between the text feature vectors is reduced. Accordingly, the classification efficiency of KNN algorithm is improved. The experimental results show that the proposed algorithm is feasible.